Legal origin and development heterogeneity: Distributional evidence from cross-country panel data

Küçükkocaoğlu G., Çakır N.

Social Sciences & Humanities Open, sa.14, ss.1-13, 2026 (Scopus)

Özet

This study investigates whether historical legal-family classification is associated with economic development in a stable manner across countries, or whether that association varies across the conditional real-income distribution. Legal origin is conceptualized as a time-invariant historical marker of institutional ancestry rather than as a direct measure of contemporary legal functioning. The analysis uses an annual panel of 140 countries over 2000–2023 and measures development with real GDP per capita in constant 2021 purchasing-power-parity international dollars. Because legal origin does not vary within countries, its coefficients are identified from between-country variation. The empirical design combines a broad parsimonious specification with Mundlak-type country means and common year effects, and evaluates sensitivity to richer conditioning sets, balanced-panel restrictions, historically overlapping country characteristics, and alternative outcome measures. Primary inference for the quantile models is based on a country-level cluster bootstrap that resamples countries and retains their full within-country observation histories. The principal estimation sample contains 139 countries and 3256 country-year observations. Mean-based estimates do not reveal a stable civil-law premium or penalty. Quantile point estimates vary economically across the conditional income distribution, but bootstrap confidence intervals are wide and formal bootstrap tests do not reject equality of the civil-law coefficient across quantiles in the principal, rich-control, or balanced specifications. Coefficient profiles nevertheless change materially with conditioning intensity, sample composition, and historically overlapping controls. The findings therefore do not support a universal legal-family effect or statistically reliable cross-quantile heterogeneity in the principal specification. In this sample, legal ancestry has no contemporary income association that can be separated from colonial history or shown to be stable across empirical designs: common-law classification coincides with English colonial-language history for 30 of 31 countries, and the estimated association moves materially with conditioning intensity and sample composition. A power calculation further indicates that a cross-country design with a time-invariant classifier cannot detect cross-quantile differences of plausible magnitude, which bears on how legal-origin estimates reported elsewhere should be read.